The Problem
The client came with a classic request: “we want a bot that answers customer questions and helps us sell more without increasing our support team. We want lead classification to be more dynamic, reaching more clients than our current team can handle.”
The core idea is simple: while the agent answers user questions, it also understands the moment and context of the user’s company, profiling the user to guide them better. The client did not want automation only; they wanted to understand who was on the other side in order to offer the right product, in the right way, at the right time.
The main pain points were:
- Lack of personalization at scale: the human team could not maintain the same level of personalization at scale, especially at peak times and outside business hours.
- Slow, manual lead classification: leads took longer to classify because the process was manual, consuming time and reducing the number of clients served quickly.
But what is an egogram?
An egogram is a Transactional Analysis tool that graphically represents how different ego states appear in a person’s behavior: the “Parent,” the “Adult,” and the “Child,” in their sub-states. In practical terms, this means communication and decision-making patterns: how someone speaks, reacts, asks, and decides during an interaction.
For business, this is extremely useful because:
- It helps identify behavioral profiles from the conversation itself, without relying only on cold forms or explicit tests. It supports adapting language, depth of explanation, argument type (security, outcomes, novelty, belonging, etc.), and even call-to-action timing.
- Instead of treating “profile” as a static label, egogram helps view the user as a set of behavioral tendencies in interaction — perfect for sales and support conversational agents.
Understanding the Agent
To meet the client’s expectation, our agent needed to understand which profiles could be used for classification, which questions to ask to extract the best information, and how to cross that data. In addition, the agent had to know when to answer specific product questions when needed.
In other words, the agent needed to:
- Answer institutional questions (who the company is, credentials, differentiators) and specific product questions.
- Profile the user during the conversation itself, without breaking flow and without sounding like a “personality test questionnaire.”
- Classify the user into a set of egogram profiles defined with the client’s team.
- Use that profile to:
- Adapt the offered product.
- Adapt the guidance given to the client.
- Send the lead, already segmented, to sales or to the next funnel step.
1. Mapping frequent questions
The first step was mapping the terrain:
- We collected the main real client questions, both institutional and product-related, from support history. At this first stage, we mapped general questions about the institution, products, catalog, and related topics.
With that, we built a conversational flow that was not just an automated FAQ, but a journey blending clarification, education, and qualification.
2. Mapping profiles
By understanding the egogram profiles defined by the client, we built a basis for how users speak, their issues and concerns, and how to direct them based on profile.
3. Conversing with the client
Egogram can be used in different ways, but here the focus was not psychological diagnosis; it was a practical reading of the client’s business moment. During the conversation, the agent should understand problems, difficulties, questions, day-to-day reality, and the overall scenario.
Instead of asking “Are you more rational or emotional?”, the agent used contextual questions:
- How the person prefers to make decisions.
- What their day-to-day difficulty is.
- How they feel about their business.
- And so on.
The model classified the user based on responses and interaction style — not only content, but form: objectivity, level of detail, level of questioning, and more. From that point on, the egogram profile became an active variable inside the flow.
4. Personalizing support and recommendations
Once the profile was identified, the agent began adapting:
- Product recommendations: the agent suggested product and plan combinations with stronger fit to the identified profile and pain.
- Guidance: it advised the client according to their profile.
At the end of the conversation, the lead was saved with:
- Egogram profile.
- Relevant information history.
- Whether they had prior contact with any product.
This completely changes the quality of contacts reaching sales.
Results observed in operations
Even without disclosing specific numbers, some effects became clear:
- Better-qualified leads: sales started receiving contacts with context and mapped profile, shortening discovery time on calls and increasing conversion rate.
- More consistent support: the agent answered frequent questions 24/7 with stable quality, reducing team overload and freeing people for complex cases.
- Personalization at scale: profiles that previously depended on each agent’s “gut feeling” started being handled systematically by the bot, with automatically adapted scripts and recommendations.
For the client, this meant less operational effort to maintain high-quality support and more intelligence in how each user is guided through the funnel. This is a case of how agent technology, egogram, and business-data integrations can be combined to solve a very human problem: how to talk to each person in the right way and offer what truly makes sense.
In the end, what we delivered was not just a bot that answers questions, but an agent that:
- Learns about the user while serving.
- Adjusts conversation to behavioral profile.
- Delivers much more qualified leads to sales.
- Works 24/7.